feat: implement Reflection agentic design pattern

[indrasn0wal]

📋 Summary

Implements the Reflection agentic design pattern (Chapter 4 of Agentic Design Patterns) in MoFA. The pattern runs a generate → critique → refine loop where an agent produces a draft, critiques it, and iteratively improves it — addressing Open Task 8.

🔗 Related Issues

Closes #179

---

🧠 Context

The Reflection pattern was missing from MoFA's agentic pattern library. While ThoughtStepType::Reflection and ThoughtStep::reflection() already existed in mofa-kernel, there was no implementation of the actual critique–refine loop. This PR fills that gap by implementing the full pattern in mofa-foundation and wiring it into the existing react module and SDK.

---

🛠️ Changes

• New reflection.rs in mofa-foundation/src/react/ — types, config, agent, builder, and run loop
• Updated react/mod.rs — registered module, exported types, added to prelude
• New examples/reflection_agent/ — end-to-end example working with any LLM provider
• Updated examples/Cargo.toml — added reflection_agent to workspace members

---

🧪 How you Tested

1. cargo test -p mofa-foundation reflection — 3 unit tests pass
2. cd examples/reflection_agent && cargo run — tested end-to-end with OpenAI GPT-4, ran 3 rounds of generate → critique → refine successfully

---

📸 Screenshots / Logs (if applicable)

(mofa-dev) indrasn0wal@Indras-MacBook-Air mofa % cargo test -p mofa-foundation reflection
   Compiling mofa-foundation v0.1.0 (/Users/indrasn0wal/Desktop/mofa/crates/mofa-foundation)
warning: cannot test inner items
   --> crates/mofa-foundation/src/collaboration/types.rs:768:9
    |
768 |         #[tokio::test]
    |         ^^^^^^^^^^^^^^
    |
    = note: #[warn(unnameable_test_items)] on by default
warning: mofa-foundation (lib test) generated 1 warning
    Finished test profile [unoptimized + debuginfo] target(s) in 9.24s
     Running unittests src/lib.rs (target/debug/deps/mofa_foundation-f1dec615a9df58fa)
running 3 tests
test react::reflection::tests::config_builder ... ok
test react::reflection::tests::config_defaults ... ok
test react::reflection::tests::reflection_step_serialization ... ok
test result: ok. 3 passed; 0 failed; 0 ignored; 0 measured; 183 filtered out; finished in 0.00s

2026-02-22T10:39:25.899747Z  INFO reflection_agent: ==========================================
2026-02-22T10:39:25.899843Z  INFO reflection_agent:   MoFA Reflection Agent Example
2026-02-22T10:39:25.899852Z  INFO reflection_agent: ==========================================
2026-02-22T10:39:25.929416Z  INFO reflection_agent: Task: Explain the concept of ownership in Rust and why it matters.

2026-02-22T10:39:33.597491Z  INFO mofa_foundation::react::reflection: [Reflection] Round 1 - critiquing draft (len=1544)
2026-02-22T10:39:40.030239Z  INFO mofa_foundation::react::reflection: [Reflection] Round 1 - refining
2026-02-22T10:39:52.784465Z  INFO mofa_foundation::react::reflection: [Reflection] Round 2 - critiquing draft (len=1927)
2026-02-22T10:39:58.277431Z  INFO mofa_foundation::react::reflection: [Reflection] Round 2 - refining
2026-02-22T10:40:08.352702Z  INFO mofa_foundation::react::reflection: [Reflection] Round 3 - critiquing draft (len=2259)
2026-02-22T10:40:15.804855Z  INFO mofa_foundation::react::reflection: [Reflection] Completed in 3 rounds, 49875ms

---

⚠️ Breaking Changes

• No breaking changes

---

🧹 Checklist

Code Quality

• Code follows Rust idioms and project conventions
• cargo fmt run
• cargo clippy passes without warnings

Testing

• Tests added/updated
• cargo test passes locally without any error

Documentation

• Public APIs documented

PR Hygiene

• PR is small and focused (one logical change)
• No unrelated commits

---

🧩 Additional Notes for Reviewers

• Critic defaults to same agent as generator — separate critic supported via .with_critic(Arc<LLMAgent>)
• Early stopping implemented — loop exits when critic says "no major improvements needed"
• Uses existing ThoughtStep::reflection() from kernel for trace consistency
• Mock LLM tests for the full run loop are left as a follow-up

[indrasn0wal]

@BH3GEI Please review it.